Evaluating and Improving the Predictive Performance of Risk Equalization Models in Health Insurance Markets

Several countries world-wide, including Belgium, Germany, Israel, the Netherlands, Switzerland, and the U.S., use a risk equalization (RE) model to provide risk-adjusted payments to health insurers. The goal of RE is to mitigate financial incentives for risk selection and thereby to achieve a level playing field for health insurers. The extent to which an RE-model achieves this goal depends on the predictive performance of this model. In contrast to the vast amount of literature paying attention to improving the predictive performance of RE-models, evaluating model’s performance has been understudied.
The first part of this thesis formulates general principles on how to evaluate model’s performance. These principles may assist researchers and policymakers by performing empirical evaluations and interpreting the results of these evaluations for decision-making. Despite RE-models have been developed over the past decades, a critical question for policymakers in all countries with RE is still how to further improve model’s predictive performance.
The second part of this thesis examines three potentially relevant methods to improve the predictive performance of sophisticated morbidity-based RE-models.